Developmental theories suggest affect-biased attention, preferential attention to emotionally salient stimuli, emerges during infancy through coordinating individual differences. Here we examined bidirectional relations between infant affect-biased attention, temperamental negative affect, and maternal anxiety symptoms using a Random Intercepts Cross-Lagged Panel model (RI-CLPM). Infant-mother pairs from Central Pennsylvania and Northern New Jersey (N = 342; 52% White; 50% reported as assigned female at birth) participated when infants were 4, 8, 12, 18 and 24 months of age. Infants completed the overlap task while eye-tracking data were collected. Mothers reported their infant's negative affect and their own anxiety symptoms. In an RI-CLPM, after accounting for between-person variance (random intercepts representing the latent average of a construct), it is possible to assess within-person variance (individual deviations from the latent average of a construct). Positive relations represent "stability" in constructs (smaller within-person deviations). Negative relations represent "fluctuation" in constructs (larger within-person deviations). At the between-person level (random intercepts), mothers with greater anxiety symptoms had infants with greater affect-biased attention. However, at the within-person level (deviations), greater fluctuation in maternal anxiety symptoms at 12- and 18 months prospectively related to greater stability in attention to angry facial configurations. Additionally, greater fluctuation in maternal anxiety symptoms at 18 months prospectively related to greater stability in attention to happy facial configurations. Finally, greater fluctuation in maternal anxiety symptoms at 4- and 12 months prospectively related to greater stability in infant negative affect. These results suggest that environmental uncertainty, linked to fluctuating maternal anxiety, may shape early socioemotional development.